Time-parameterized queries in spatio-temporal databases

TLDR

Time‑parameterized queries retrieve the current result, its validity period, and the change that causes expiration, and are crucial for highly dynamic spatio‑temporal applications, yet efficient processing remains largely unexplored. The authors propose a general framework that unifies time‑parameterized window, k‑nearest neighbor, and spatial join queries. Each TP query is reduced to a nearest‑neighbor search with query‑specific distance functions, enabling application to mobile queries or objects and evaluation using R‑trees and their dynamic extensions. This reduction permits the use of well‑known branch‑and‑bound techniques, improving efficient processing of time‑parameterized queries.

Abstract

Time-parameterized queries (TP queries for short) retrieve (i) the actual result at the time that the query is issued, (ii) the validity period of the result given the current motion of the query and the database objects, and (iii) the change that causes the expiration of the result. Due to the highly dynamic nature of several spatio-temporal applications, TP queries are important both as standalone methods, as well as building blocks of more complex operations. However, little work has been done towards their efficient processing. In this paper, we propose a general framework that covers time-parameterized variations of the most common spatial queries, namely window queries, k-nearest neighbors and spatial joins. In particular, each of these TP queries is reduced to nearest neighbor search where the distance functions are defined according to the query type. This reduction allows the application and extension of well-known branch and bound techniques to the current problem. The proposed methods can be applied with mobile queries, mobile objects or both, given a suitable indexing method. Our experimental evaluation is based on R-trees and their extensions for dynamic objects.